A photovoltaic control parameter online correction method, system, device and medium

Abstract

The application relates to a photovoltaic control parameter online correction method, system, device and medium, the method comprising: acquiring real-time operation data of a photovoltaic module, inputting a composite photovoltaic model based on double-time-scale extended Kalman filtering, obtaining a double-physical-field decoupling state parameter set and a state covariance matrix; performing physical consistency verification on the double-physical-field decoupling state parameter set, triggering local optimization correction under physical constraints when a rule is violated, obtaining a correction parameter set, and combining current environmental data to obtain a theoretical maximum power point voltage through model predictive control operation, thereby generating a feedforward duty cycle instruction; generating a feedback correction instruction based on the state covariance matrix and the environmental change rate, and superimposing the two to obtain a final control instruction. The method can fuse real-time operation state and dynamic environmental information of the photovoltaic module, realize stable tracking of the maximum power point of the photovoltaic module under complex working conditions, and effectively improve the energy conversion efficiency and operation stability of the photovoltaic power generation system.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic power generation technology, and in particular relates to a method, system, equipment and medium for online correction of photovoltaic control parameters. Background Technology

[0002] The installed capacity of large-scale centralized photovoltaic power plants and distributed photovoltaic systems continues to climb. Against this industry backdrop, ensuring the efficient and stable operation of photovoltaic power generation systems throughout their entire lifecycle, and maximizing the energy conversion efficiency and overall power generation revenue of photovoltaic modules, has become a core research direction in this field. Consequently, online identification and correction technologies for photovoltaic module control parameters have been widely applied and developed. These technologies can obtain real-time operating data from photovoltaic modules and correct core electrical parameters characterizing module output features online. This provides accurate model support for maximum power point tracking (MPPT) control and data-driven data for photovoltaic module health status assessment and power plant operation and maintenance decisions.

[0003] In traditional technologies, online calibration of photovoltaic module control parameters often employs simplified single-domain photovoltaic equivalent circuit models, combined with conventional filtering identification algorithms, disturbance observation methods, and incremental conductance methods, to perform online fitting and compensation for parameter drift during photovoltaic module operation. Some solutions introduce environmental parameter correction factors to adapt to fluctuations in module output characteristics caused by changes in temperature and irradiance, thereby achieving online calibration of control parameters and ensuring the basic control effect of the MPPT algorithm.

[0004] However, current online calibration methods for photovoltaic (PV) control parameters still have unavoidable technical flaws. During long-term operation, the performance of the core component of a PV power generation system—the PV module—is not static. First, with accumulated operating time, the internal materials of the module undergo irreversible aging. The electrical characteristics of this aging are mainly manifested in a slow increase in series resistance, leading to a decrease in output power. Second, ambient temperature dynamically affects module performance in real time, especially exhibiting rapid drift in parameters such as parallel resistance. Traditional online parameter calibration methods, in pursuit of computational speed, often employ simplified single-parameter models or fail to accurately distinguish the effects of these two physical mechanisms. When modules simultaneously experience significant aging and drastic temperature fluctuations, these methods are prone to parameter compensation misjudgments—incorrectly attributing power loss due to aging to temperature changes or misjudging temperature effects as aging. This misjudgment not only leads to inaccurate short-term power predictions but, more seriously, masks the true health status, causing maintenance personnel to miss optimal maintenance opportunities. It also causes maximum power point tracking algorithms based on these erroneous parameters to operate in a suboptimal state for extended periods, resulting in considerable economic losses. Summary of the Invention

[0005] Therefore, it is necessary to provide a photovoltaic control parameter online correction method, system, equipment, and medium that can integrate the real-time operating status and dynamic environmental information of photovoltaic modules to achieve stable tracking of the maximum power point of photovoltaics under complex operating conditions, in order to address the above-mentioned technical problems.

[0006] Firstly, this application provides a method for online calibration of photovoltaic control parameters, including:

[0007] S101. Obtain the real-time operating data of the photovoltaic module, input the real-time operating data into the composite photovoltaic model based on the dual-time-scale extended Kalman filter mechanism, and obtain the dual-physics decoupled state parameter set and state covariance matrix.

[0008] S102. Perform physical consistency verification on each parameter of the dual-physics decoupling state parameter set. When any parameter in the dual-physics decoupling state parameter set violates the preset physical constraint rules, trigger local optimization correction under physical constraints to obtain the correction parameter set.

[0009] S103. Obtain the environmental data at the current moment, perform model predictive control calculations on the correction parameter set and environmental data, and obtain the theoretical maximum power point voltage; wherein, the theoretical maximum power point voltage is used to characterize the ideal operating voltage of the photovoltaic module when it outputs maximum power under the current environmental conditions;

[0010] S104. Generate a feedforward duty cycle command based on the theoretical maximum power point voltage; wherein, the feedforward duty cycle command is used to instruct the inverter to control the operating voltage of the photovoltaic module to approach the theoretical maximum power point voltage.

[0011] S105. Perform differential calculations on real-time operating data and environmental data to obtain the environmental change rate. Based on the state covariance matrix and the environmental change rate, generate a feedback correction instruction and superimpose the feedforward duty cycle instruction and the feedback correction instruction to obtain the final control instruction. Among them, the feedback correction instruction is used to instruct the inverter to adjust the operating voltage of the photovoltaic module; the final control instruction is used to instruct the inverter to adjust the on-duty cycle of the switching transistor.

[0012] Secondly, this application also provides an online correction system for photovoltaic control parameters, comprising:

[0013] The parameter decoupling module is used to acquire real-time operating data of photovoltaic modules. The real-time operating data is input into a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism to obtain a dual-physics decoupling state parameter set and a state covariance matrix.

[0014] The optimization and correction module is used to perform physical consistency verification on each parameter of the dual-physics decoupling state parameter set. When any parameter in the dual-physics decoupling state parameter set violates the preset physical constraint rules, local optimization and correction under physical constraints is triggered to obtain the correction parameter set.

[0015] The predictive control module is used to acquire the environmental data at the current moment, perform model predictive control calculations on the correction parameter set and environmental data, and obtain the theoretical maximum power point voltage; wherein, the theoretical maximum power point voltage is used to characterize the ideal operating voltage of the photovoltaic module when it outputs maximum power under the current environmental conditions;

[0016] The feedforward control module is used to generate a feedforward duty cycle command based on the theoretical maximum power point voltage; the feedforward duty cycle command is used to instruct the inverter to control the operating voltage of the photovoltaic module to approach the theoretical maximum power point voltage.

[0017] The feedback control module performs differential calculations on real-time operating data and environmental data to obtain the environmental change rate. Based on the state covariance matrix and the environmental change rate, it generates feedback correction instructions and superimposes the feedforward duty cycle instructions and feedback correction instructions to obtain the final control instructions. Among them, the feedback correction instructions are used to instruct the inverter to adjust the operating voltage of the photovoltaic modules; the final control instructions are used to instruct the inverter to adjust the on-duty cycle of the switching transistors.

[0018] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described above.

[0019] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0020] The aforementioned online correction method, system, device, and medium for photovoltaic control parameters, by acquiring real-time operating data of photovoltaic modules and performing decoupling calculations using a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism, can extract the core state parameters and state covariance matrix of the photovoltaic module's dual-physics field. This achieves high-precision, high-dynamic decoupling characterization of the photovoltaic module's operating state, laying a data foundation for subsequent state correction and power point tracking. By performing full-parameter physical consistency verification on the decoupled state parameter set of the dual-physics field and performing local optimization correction under physical constraints for parameters that violate preset physical constraint rules, outliers and physical inconsistencies in the parameter calculation process can be effectively eliminated, ensuring the physical rationality and reliability of the state parameters and fundamentally avoiding maximum power point tracking deviations caused by parameter distortion. By combining current environmental data and the correction parameter set for model predictive control calculations, the theoretical maximum power point voltage of the photovoltaic module under the current environmental conditions is solved, and a feedforward duty cycle command is generated based on this voltage. This enables control of the photovoltaic module based on... This approach utilizes real-time operating characteristics and environmental conditions to achieve feedforward prediction of the maximum power point (MPPT), improving the response speed of MPPT and avoiding the lag issues of traditional tracking algorithms. By performing differential operations on real-time operating and environmental data to obtain the environmental change rate, and combining the state covariance matrix with the environmental change rate to generate adaptive feedback correction commands, the feedforward duty cycle command and feedback correction command are superimposed to generate the final control command. This enables closed-loop adaptive correction based on dynamic environmental changes and state estimation uncertainties, balancing the speed and steady-state accuracy of the photovoltaic power tracking process. Thus, it achieves a closed-loop technology for the entire photovoltaic MPPT process, from high-precision decoupling of photovoltaic module operating status, parameter correction under physical constraints, MPPT feedforward prediction to environmental adaptive feedback correction. This effectively solves the core pain points of traditional MPPT algorithms, such as lag response, steady-state oscillation, and insufficient tracking accuracy in complex environments, significantly improving the dynamic response speed, steady-state control accuracy, and adaptability to complex operating conditions of photovoltaic module MPPT. This method integrates real-time operating status and dynamic environmental information of photovoltaic modules to achieve stable tracking of the maximum power point of photovoltaics under complex operating conditions, effectively improving the energy conversion efficiency and operational stability of photovoltaic power generation systems. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1A schematic flowchart illustrating an online correction method for photovoltaic control parameters, provided as an exemplary embodiment of this application;

[0023] Figure 2 This is a schematic diagram of the structure of an online photovoltaic control parameter correction system provided as an exemplary embodiment of this application. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0025] In one embodiment, such as Figure 1 As shown, an online calibration method for photovoltaic control parameters is provided. This embodiment illustrates the application of this method to an intelligent calibration terminal. It is understood that this method can also be applied to an intelligent calibration server, and further to a system including both an intelligent calibration terminal and an intelligent calibration server, and is implemented through the interaction between the intelligent calibration terminal and the intelligent calibration server. In this embodiment, the method may include the following steps:

[0026] S101. Obtain real-time operating data of photovoltaic modules, input the real-time operating data into a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism, and obtain the dual-physics decoupled state parameter set and state covariance matrix.

[0027] Optionally, the real-time operating data of the photovoltaic module can be used to characterize the photovoltaic module during real-time operation. The data collected by the sampling circuit and sensors can reflect the electrical output characteristics and operating status of the module. The real-time operating data of the photovoltaic module may include, but is not limited to, the real-time port voltage, real-time output current, internal temperature of the cell and real-time irradiance of the module surface.

[0028] Optionally, the dual-physics decoupling state parameter set can be used to characterize the set of core electrical characteristic parameters that correspond to the slow-changing aging physical field and the fast-changing environmental physical field of the photovoltaic module, obtained after decoupling calculation through the composite photovoltaic model. The dual-physics decoupling state parameter set may include, but is not limited to, parameters such as series resistance, aging rate coefficient, photocurrent and parallel resistance.

[0029] Optionally, the state covariance matrix can be used to characterize the error covariance matrix between the estimated and true values ​​of each state parameter during the state estimation process. The state covariance matrix is ​​used to quantify the estimation uncertainty of each state parameter.

[0030] Specifically, the intelligent calibration terminal can be adapted to the sampling communication interface of the photovoltaic inverter and the temperature and irradiance sensors deployed on the module side. It synchronously collects real-time operating data of the photovoltaic modules according to a preset sampling period, performs moving average filtering preprocessing on the collected raw data, and removes impulse noise and random interference during the sampling process to obtain standardized input data. Furthermore, the intelligent calibration terminal can input the preprocessed real-time operating data into a pre-constructed composite photovoltaic model. The composite photovoltaic model can be built based on a dual-timescale extended Kalman filter mechanism. The composite photovoltaic model can set a large timescale filtering step size for the slow-changing aging process and a small timescale filtering step size for rapid environmental fluctuations. Through dual-timescale prediction and iterative update calculations, the composite photovoltaic model decouples the parameters of the slow-changing aging physical field and the rapid-changing environmental physical field, outputting a dual-physical-field decoupled state parameter set and synchronously outputting the iteratively updated state covariance matrix.

[0031] S102. Perform physical consistency verification on each parameter of the dual-physics decoupling state parameter set. When any parameter in the dual-physics decoupling state parameter set violates the preset physical constraint rules, trigger local optimization correction under physical constraints to obtain the correction parameter set.

[0032] Optionally, the preset physical constraint rules can be multi-dimensional constraint rules that are pre-set based on the electrical principles of photovoltaic modules, the physical characteristics of semiconductors and engineering operation experience, and are used to determine the physical rationality of state parameters. The physical constraint rules may include, but are not limited to, parameter physical boundary constraint rules, series resistance monotonicity constraint rules, aging rate coefficient change rate constraint rules and physical dependency relationship constraint rules between parameters.

[0033] Optionally, the set of correction parameters can be used to characterize the set of core state parameters of a photovoltaic module that meets the requirements of data fitting accuracy and physical consistency after being obtained by local optimization correction under physical constraints for decoupled state parameters that violate physical constraint rules.

[0034] Specifically, the intelligent calibration terminal can load pre-stored physical constraint rules matched to the photovoltaic module model. It can then perform multi-dimensional physical consistency checks on each parameter in the output dual-physical field decoupling state parameter set, determining whether the parameter exceeds a preset physical boundary range, whether the series resistance conforms to the physical characteristic of monotonically increasing with operating time, whether the fluctuation of the aging rate coefficient exceeds a preset reasonable range, and whether the physical dependencies between parameters conform to the basic principles of the photovoltaic equivalent circuit. When any parameter fails the check, the intelligent calibration terminal can trigger local optimization correction under physical constraints. Using the weighted sum of data fitting error and physical constraint violation penalty as the objective function, and the dual-physical field decoupling state parameter set as the initial iteration value, it searches for the optimal solution where the objective function converges within the physical boundary threshold range using a gradient descent algorithm. The parameter set corresponding to the optimal solution is then used as the calibration parameter set.

[0035] S103. Obtain the environmental data at the current moment, perform model predictive control calculations on the correction parameter set and environmental data, and obtain the theoretical maximum power point voltage.

[0036] Optionally, the environmental data at the current moment may include, but is not limited to, the real-time ambient temperature at the location, the normal irradiance of the component surface, wind speed, atmospheric pressure, and relative humidity.

[0037] Alternatively, the theoretical maximum power point voltage can be used to characterize the ideal operating voltage of a photovoltaic module when it outputs maximum power under current environmental conditions.

[0038] Specifically, the intelligent correction terminal can obtain the current environmental data through the environmental sensor group and the meteorological data communication interface. After removing outliers and smoothing the environmental data, it extracts the irradiance and temperature data that are strongly correlated with the output characteristics of the photovoltaic module as the core environmental input.

[0039] Furthermore, the intelligent calibration terminal can use the calibration parameter set as the real parameters of the equivalent circuit model of the photovoltaic module. Combined with the core environmental data at the current moment, within the preset control time domain, with the optimization objective of maximizing the output power of the photovoltaic module and the operating voltage safety range of the photovoltaic module as the constraint, it performs rolling optimization calculations. By traversing and solving the extreme points of the output power of the photovoltaic module with respect to the operating voltage, it obtains the theoretical maximum power point voltage corresponding to the maximum output power of the photovoltaic module under the current environmental conditions.

[0040] S104. Generate feedforward duty cycle command based on theoretical maximum power point voltage.

[0041] Among them, the feedforward duty cycle command can be used to instruct the inverter to control the operating voltage of the photovoltaic module to approach the theoretical maximum power point voltage.

[0042] Specifically, the intelligent correction terminal can load the pre-stored topology parameters and circuit transfer function of the photovoltaic module DC side to the inverter main circuit, and calculate the voltage tracking target difference based on the obtained theoretical maximum power point voltage and the current real-time port voltage of the photovoltaic module.

[0043] Furthermore, the intelligent correction terminal can be based on the control principle of pulse width modulation (PWM) of the inverter, combined with the mapping relationship between the DC bus voltage and the theoretical maximum power point voltage, and convert the theoretical maximum power point voltage into an initial duty cycle control quantity through a pre-calibrated voltage-duty cycle conversion function. It can also dynamically compensate for the initial duty cycle control quantity by combining the circuit transfer function, thereby eliminating the control deviation caused by the parasitic parameters of the main circuit and generating a feedforward duty cycle command.

[0044] S105. Perform differential operations on real-time operating data and environmental data to obtain the environmental change rate. Based on the state covariance matrix and the environmental change rate, generate feedback correction instructions and superimpose feedforward duty cycle instructions and feedback correction instructions to obtain the final control instructions.

[0045] Optionally, the feedback correction command is used to instruct the inverter to adjust the operating voltage of the photovoltaic modules.

[0046] Optionally, the final control command is used to instruct the inverter to adjust the duty cycle of the switching transistors.

[0047] Specifically, the intelligent correction terminal can perform first-order differential operations on continuously collected real-time operating data and environmental data based on a preset sampling time interval, and calculate the irradiance change rate, temperature change rate and output power change rate respectively. The irradiance change rate, temperature change rate and output power change rate are then weighted and fused to obtain the comprehensive environmental change rate.

[0048] Furthermore, the intelligent correction terminal can extract the estimated error variance of each state parameter from the state covariance matrix to obtain the state uncertainty index. Combining the environmental change rate and the state uncertainty index, the adaptive feedback correction gain is calculated through a nonlinear mapping function.

[0049] Furthermore, the intelligent correction terminal can calculate the voltage tracking error between the theoretical maximum power point voltage and the real-time port voltage of the photovoltaic module, and generate feedback correction commands by combining the adaptive feedback correction gain and the environmental rate of change. The intelligent correction terminal linearly superimposes the feedforward duty cycle command and the feedback correction command, performs amplitude limiting on the superimposed control quantity, and generates the final control command.

[0050] In the aforementioned online calibration method for photovoltaic control parameters, the intelligent calibration terminal acquires real-time operating data of the photovoltaic module and performs decoupling calculations using a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism. This allows for the extraction of core state parameters and state covariance matrices of the photovoltaic module's dual-physics field, achieving high-precision and high-dynamic decoupling characterization of the photovoltaic module's operating state and laying a data foundation for subsequent state calibration and power point tracking. By performing full-parameter physical consistency verification on the decoupled state parameter set of the dual-physics field and executing local optimization correction under physical constraints for parameters that violate preset physical constraint rules, outliers and physical inconsistencies in the parameter calculation process can be effectively eliminated, ensuring the physical rationality and reliability of the state parameters and fundamentally avoiding maximum power point tracking deviations caused by parameter distortion. By combining current environmental data and the calibration parameter set for model predictive control calculations, the theoretical maximum power point voltage of the photovoltaic module under the current environmental conditions is solved, and a feedforward duty cycle command is generated based on this voltage. This enables the control of the photovoltaic module based on its operating parameters. By using real-time operating characteristics and environmental conditions to achieve feedforward prediction of the maximum power point (MPPT), the response speed of MPPT is improved, avoiding the lag defects of traditional tracking algorithms. The environmental change rate is obtained by performing differential operations on real-time operating data and environmental data. An adaptive feedback correction command is generated by combining the state covariance matrix and the environmental change rate. The feedforward duty cycle command and the feedback correction command are then superimposed to generate the final control command. Based on rapid feedforward tracking, closed-loop adaptive correction is achieved according to dynamic environmental changes and state estimation uncertainties, balancing the speed and steady-state accuracy of the photovoltaic power tracking process. This enables a closed-loop technology for the entire photovoltaic MPPT process, from high-precision decoupling of photovoltaic module operating status, parameter correction under physical constraints, feedforward prediction of the MPPT, to adaptive feedback correction of the environment. This effectively solves the core pain points of traditional MPPT algorithms, such as lag in response, steady-state oscillation, and insufficient tracking accuracy in complex environments, significantly improving the dynamic response speed, steady-state control accuracy, and adaptability to complex operating conditions of photovoltaic module MPPT. This method integrates real-time operating status and dynamic environmental information of photovoltaic modules to achieve stable tracking of the maximum power point of photovoltaics under complex operating conditions, effectively improving the energy conversion efficiency and operational stability of photovoltaic power generation systems.

[0051] In one embodiment, real-time operating data is input into a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism to obtain a dual-physics decoupled state parameter set and a state covariance matrix, which may include:

[0052] S201. Through the slow-changing aging sub-model layer, the port voltage, cell temperature and irradiance are extracted from the real-time operating data to construct the input vector. The output current is extracted from the real-time operating data to construct the observation vector, and the state vector is defined.

[0053] Optionally, the composite photovoltaic model may include a slow-changing aging sub-model layer, a fast-changing environment sub-model layer, a dual-timescale extended Kalman filter prediction layer, a dual-timescale extended Kalman filter update layer, and an output layer.

[0054] Optionally, the state vector may include, but is not limited to, series resistance, aging rate coefficient, photocurrent, and parallel resistance.

[0055] For example, the intelligent correction terminal can complete the construction of the model input vector, observation vector and definition of the state vector through the slow aging sub-model layer of the composite photovoltaic model.

[0056] Furthermore, the slow-varying aging sub-model layer can extract the core boundary parameters that determine the output characteristics of the photovoltaic module, namely the photovoltaic module port voltage, cell temperature, and irradiance, from the real-time operating data at the same sampling time, based on the physical characteristics of the photovoltaic single-diode equivalent circuit. These parameters are then standardized according to a preset fixed dimension and arrangement rule to construct an input vector that perfectly matches the dimension of subsequent filtering operations. Further, the slow-varying aging sub-model layer can extract the output current of the photovoltaic module at the same sampling time. The output current can be the directly observable output response of the photovoltaic module under the corresponding input boundary. Based on the output current of the photovoltaic module at the same sampling time, the slow-varying aging sub-model layer can construct a single-dimensional standardized observation vector, which serves as the observation benchmark for residual calculation in the subsequent filtering process. Even further, the slow-varying aging sub-model layer can define a state vector based on the dual-physics field decoupling requirement.

[0057] S202. Through the rapidly changing environment sub-model layer, based on the cell temperature and irradiance, combined with the preset mapping relationship between the photogenerated current, parallel resistance and environmental factors, the first environmental estimate and the second environmental estimate are calculated.

[0058] Optionally, the mapping relationship between the photocurrent, parallel resistance, and environmental factors can be two sets of independent nonlinear mapping functions. The first set of nonlinear mapping functions can be a mapping function between the photocurrent and irradiance and cell temperature. This first set of nonlinear mapping functions can be constructed based on the physical mechanism of the photovoltaic effect of photovoltaic devices, using the reference value of the photocurrent under standard test conditions as a basis, combined with the linear response characteristics of irradiance and the temperature compensation coefficient of cell temperature. The second set of nonlinear mapping functions can be a mapping function between the parallel resistance and irradiance and cell temperature. It is obtained by fitting multi-condition environmental simulation test data based on the environmental response characteristics of photovoltaic cell leakage current. Both sets of mapping relationships are pre-fixed in the composite photovoltaic model and can be directly called online.

[0059] Optionally, the first environmental estimate is used to characterize the environmental estimate of the photogenerated current; the second environmental estimate is used to characterize the environmental estimate of the parallel resistance.

[0060] Optionally, the fast-changing environment sub-model layer of the composite photovoltaic model can be set based on the environmental response characteristics of photovoltaic modules. Environmental parameters such as irradiance and cell temperature fluctuate rapidly with meteorological conditions and shading, directly driving the rapid changes in photocurrent and parallel resistance. This enables the pre-estimation of fast-changing parameters and achieves time-scale matching with slow-changing aging parameters.

[0061] For example, when the intelligent correction terminal receives cell temperature and irradiance data at the same sampling time through the fast-changing environment sub-model layer of the composite photovoltaic model, it substitutes the cell temperature and irradiance data at the same sampling time into a preset mapping function to calculate a first environmental estimate representing the environmental estimate of the photogenerated current and a second environmental estimate representing the environmental estimate of the parallel resistance.

[0062] S203. By extending the Kalman filter prediction layer through dual time scales, state prediction is performed on the series resistance and aging rate coefficient to obtain the first prediction value and the second prediction value. The first prediction value, the second prediction value, the first environmental estimate value and the second environmental estimate value are fused to obtain the predicted state vector.

[0063] Optionally, the first predicted value can be used to characterize the predicted value of the series resistance.

[0064] Optionally, the second predicted value can be used to characterize the predicted value of the aging rate coefficient.

[0065] For example, the intelligent correction terminal can perform prior prediction and fusion of all state parameters through the dual-timescale extended Kalman filter prediction layer of the composite photovoltaic model.

[0066] Preferably, the Extended Kalman Filter (EKF) can be a state-optimal estimation algorithm for nonlinear systems, which achieves recursive optimal estimation of the system state by linearizing the nonlinear system through a first-order Taylor expansion.

[0067] Furthermore, the dual-timescale extended Kalman filter prediction layer can predict the current state based on the corrected state parameters and state covariance matrix from the previous sampling time, combined with a predefined slow-variable parameter state transition equation, for the slow-timescale series resistance and aging rate coefficient in the state vector. This yields a first predicted value for the series resistance and a second predicted value for the aging rate coefficient. The dual-timescale extended Kalman filter prediction layer can then fuse the first and second predicted values ​​with the first and second environment estimates to obtain the predicted state vector for the current sampling time.

[0068] S204. By extending the Kalman filter update layer through dual time scales, the predicted output current is obtained based on the predicted state vector and the input vector. The Kalman gain is obtained by calculating the residual between the predicted output current and the observation vector. The predicted state vector is corrected according to the Kalman gain to obtain the corrected state vector, and the state covariance matrix is ​​updated.

[0069] Optionally, the corrected state vector can be used to characterize the theoretical output current value calculated by the nonlinear observation equation in the composite photovoltaic model based on the current predicted state vector and the input vector. The corrected state vector can be used to perform residual calculation with the measured output current.

[0070] Optionally, the predicted output current can be used to characterize the physical mapping relationship between the input parameters and output current of the photovoltaic module, and to adapt to the nonlinear system processing requirements of the extended Kalman filter.

[0071] Optionally, the residual can be the difference between the predicted value and the actual observed value, and the residual can be used to characterize the deviation between the prior estimation result and the actual operating state.

[0072] For example, the intelligent correction terminal can use the dual-timescale extended Kalman filter update layer of the composite photovoltaic model to substitute the predicted state vector and input vector obtained in the aforementioned steps into the nonlinear observation equation constructed based on the single diode equivalent circuit of the photovoltaic device, and calculate the predicted output current at the current sampling time.

[0073] Furthermore, the dual-timescale extended Kalman filter update layer can calculate the residual between the predicted output current and the observation vector. Based on the gain update equation of the dual-timescale extended Kalman filter, and combining the prior value of the current state covariance matrix with the statistical characteristics of the observation noise, the dual-timescale extended Kalman filter update layer calculates the Kalman gain at the current time. Based on the Kalman gain and the residual, the predicted state vector is corrected to obtain the corrected state vector at the current time, thus completing the update of the state covariance matrix and obtaining the posterior state covariance matrix at the current time.

[0074] S205. Through the output layer, the corrected state vector is output as the set of state parameters for the two-physics field decoupling, and the state covariance matrix is ​​output.

[0075] For example, the intelligent correction terminal can output the dual-physics decoupled state parameter set and state covariance matrix through the output layer of the composite photovoltaic model.

[0076] Optionally, the dual physical fields can be the aging physical field of the photovoltaic device and the environmental disturbance physical field, respectively. The aging physical field corresponds to the material aging and performance degradation process throughout the entire life cycle of the photovoltaic device, and the characteristic parameters of the aging physical field may include, but are not limited to, slow timescale parameters. The environmental disturbance physical field corresponds to the changes in the external environment such as weather and sunlight during the operation of the photovoltaic module, and the characteristic parameters of the environmental disturbance physical field may include, but are not limited to, fast timescale parameters.

[0077] For example, the output layer of the composite photovoltaic model can output the modified state vector as a set of decoupled state parameters of the two physics fields, and simultaneously output the updated state covariance matrix.

[0078] In this embodiment, the intelligent correction terminal constructs a composite photovoltaic model and integrates the prediction and update mechanism of dual-timescale extended Kalman filtering to achieve online decoupling and high-precision collaborative estimation of state parameters for two types of physical fields: photovoltaic module aging and environmental disturbance. This provides a parameter foundation with strong physical consistency and good dynamic adaptability for subsequent maximum power point tracking.

[0079] In one embodiment, the method may further include the following steps:

[0080] S301. Obtain the historical calibration parameter set and the corresponding historical environmental data after physical consistency verification. Use the series resistance sequence in the historical calibration parameter set as the dependent variable and the cumulative irradiance and cumulative temperature stress in the historical environmental data as independent variables to construct an aging characteristic fitting model. Solve the coefficients of the aging characteristic fitting model through an offline regression algorithm to obtain the updated aging rate parameters.

[0081] Optionally, historical environmental data can be used to characterize the cumulative environmental stress monitoring data throughout the entire operating cycle of photovoltaic modules. Historical environmental data may include, but is not limited to, the two core factors driving device aging: cumulative irradiance and cumulative temperature stress.

[0082] Optionally, the aging characteristic fitting model can be used to characterize the quantitative mapping relationship between the aging process of photovoltaic devices and the cumulative amount of environmental stress.

[0083] For example, the intelligent calibration terminal can acquire a set of historical calibration parameters that has undergone physical consistency verification, as well as historical environmental data that strictly corresponds to the time sequence of the historical calibration parameter set. The intelligent calibration terminal can use the series resistance sequence in the historical calibration parameter set as the dependent variable; the intelligent calibration terminal can use the cumulative irradiance and cumulative temperature stress in the historical environmental data as independent variables, and construct an aging characteristic fitting model based on the aging mechanism of photovoltaic devices.

[0084] Furthermore, the intelligent correction terminal can solve the coefficients of the aging characteristic fitting model through a predefined offline regression algorithm. The offline regression algorithm can perform global fitting based on a large amount of historical data, effectively avoiding the limitations of real-time online calculation, improving the accuracy of coefficient solution, and obtaining the updated aging rate parameters through the solution.

[0085] S302. Synchronize the updated aging rate parameters to the slow-varying aging sub-model layer to update the slow-varying aging sub-model layer.

[0086] For example, the intelligent calibration terminal can synchronize the updated aging rate parameters to the slow-varying aging sub-model layer through a preset internal data transmission link. During transmission, the intelligent calibration terminal can employ a data verification mechanism to prevent parameter transmission distortion or loss. After data synchronization is complete, the slow-varying aging sub-model layer can replace the original aging rate coefficient baseline value.

[0087] In this embodiment, the intelligent calibration terminal constructs an aging feature fitting model based on historical calibration parameter sets and environmental data, and then updates the aging rate parameters using an offline regression algorithm before synchronizing it to the slow-changing aging sub-model layer. This enables adaptive modeling of the aging process of photovoltaic modules and continuous optimization of model parameters, thereby improving the parameter accuracy and long-term prediction capability of the composite photovoltaic model throughout its entire life cycle.

[0088] In one embodiment, generating feedback correction instructions based on the state covariance matrix and the rate of change of the environment may include the following steps:

[0089] S401. Extract the diagonal elements from the state covariance matrix to construct the state estimation error variance vector. Calculate the norm of the state estimation error variance vector and use it as the state uncertainty index. Normalize the environmental change rate and calculate the modulus to obtain the environmental dynamic disturbance intensity index.

[0090] Optionally, the state estimation error variance vector can be a set of individual estimation error variances for each state parameter. The state estimation error variance vector can be used to characterize the dispersion and uncertainty of the estimation results for a single state parameter.

[0091] Optionally, the modulus can be used to characterize the overall intensity of the rate of environmental change.

[0092] Optionally, the norm of the state estimation error variance vector can comprehensively characterize the overall uncertainty of all state parameter estimation results.

[0093] Optionally, the environmental dynamic disturbance intensity index can be used to characterize the degree of drastic change in the environment in which the photovoltaic module is currently located.

[0094] For example, the intelligent correction terminal can extract all diagonal elements from the state covariance matrix to construct a state estimation error variance vector. The intelligent correction terminal can calculate the norm of the state estimation error variance vector and use it as a state uncertainty index. Furthermore, the intelligent correction terminal can acquire the time-series changes in real-time environmental data, calculate the environmental change rate, normalize the environmental change rate to eliminate dimensional differences between different environmental parameters, and obtain an environmental dynamic disturbance intensity index by calculating the modulus of the normalized environmental change rate.

[0095] S402. Perform nonlinear mapping on the state uncertainty index and the environmental dynamic disturbance intensity index to obtain the feedback correction gain.

[0096] Among them, the feedback correction gain can be used to characterize the inverter's response strength to voltage deviation.

[0097] For example, the intelligent correction terminal can obtain a feedback correction gain that is adapted to the current operating conditions of the photovoltaic system by using a predefined nonlinear mapping operation based on the state uncertainty index and the environmental dynamic disturbance intensity index generated in the aforementioned steps.

[0098] Optionally, the predefined nonlinear mapping function can be pre-constructed based on the closed-loop control stability and dynamic response characteristics of the photovoltaic system. The predefined nonlinear mapping function can adaptively adjust the gain according to the changes in the state uncertainty index and the environmental dynamic disturbance intensity index.

[0099] S403. Calculate the voltage tracking error between the theoretical maximum power point voltage and the real-time port voltage of the photovoltaic module. Use the product of the voltage tracking error and the feedback correction gain as the proportional correction term, and define the environmental change rate as the feedforward compensation term. Linearly superimpose the proportional correction term and the feedforward compensation term to obtain the feedback correction command.

[0100] Optionally, voltage tracking error can be used to characterize the degree of deviation between the actual operating voltage of the photovoltaic module and the ideal maximum power point operating voltage under current operating conditions.

[0101] For example, the intelligent correction terminal can calculate the difference between the theoretical maximum power point voltage and the real-time port voltage based on the previously generated theoretical maximum power point voltage and the real-time port voltage extracted from the real-time operating data synchronously collected from the photovoltaic module, thus obtaining the voltage tracking error. Further, the intelligent correction terminal can multiply the voltage tracking error by the adaptive feedback correction gain generated in the aforementioned steps to obtain a proportional correction term. The intelligent correction terminal can define the environmental change rate obtained through the aforementioned differential operation as a feedforward compensation term, used to compensate for voltage deviations caused by dynamic environmental fluctuations in advance, and to compensate for the response lag of the proportional correction stage. The proportional correction term and the feedforward compensation term are linearly superimposed and processed through a preset safety range limit to obtain a feedback correction command.

[0102] In this embodiment, the intelligent correction terminal extracts the state uncertainty index from the state covariance matrix and constructs the environmental dynamic disturbance intensity index by combining it with the environmental change rate. After nonlinear mapping, it obtains the adaptive feedback correction gain. Then, it linearly superimposes the proportional correction term of the voltage tracking error with the feedforward compensation term of the environmental change rate to generate the feedback correction command. This enables the feedback gain to be adaptively adjusted according to the state estimation uncertainty and the intensity of environmental disturbance, thereby improving the dynamic response speed and steady-state tracking accuracy of the photovoltaic system under complex operating conditions.

[0103] In one embodiment, physical consistency verification is performed on each parameter of the dual-physics decoupling state parameter set. When any parameter in the dual-physics decoupling state parameter set violates a preset physical constraint rule, local optimization correction under physical constraints is triggered to obtain a correction parameter set, which may include the following steps:

[0104] S501. Perform multi-dimensional physical consistency verification on the decoupled state parameter set of the dual physics field to obtain multiple verification results. When any of the multiple verification results is determined to be unsuccessful, trigger local optimization correction under physical constraints to construct the objective function.

[0105] Alternatively, the objective function can be used to characterize a weighted sum of data fitting error and physical constraint violation penalties.

[0106] Preferably, the expression for the objective function can be:

[0107]

[0108] In the formula, Describe the objective function. This represents the measured output current. This represents the set of parameters to be optimized. Represents the set of parameters to be optimized. The output current of the model is obtained by substituting into the composite photovoltaic model. This represents the boundary constraint penalty term. This represents the monotonicity penalty term for the series resistance. This represents the penalty term for the rate of change of the aging rate coefficient. This represents the penalty weight coefficient, and .

[0109] For example, the intelligent calibration terminal can perform multi-dimensional physical consistency verification on the decoupled state parameter set of dual physical fields. The verification dimensions can be pre-set based on the semiconductor physical characteristics of photovoltaic devices, the equivalent circuit principle of a single diode, and the operating rules throughout the entire life cycle. The verification dimensions may include, but are not limited to, parameter physical boundary constraints, series resistance monotonicity constraints, aging rate coefficient change rate constraints, and physical dependency constraints between parameters. Furthermore, the intelligent calibration terminal can output multiple verification results corresponding to each constraint after completing the verification for each constraint.

[0110] Furthermore, when any one of the multiple verification results is deemed unsuccessful, the intelligent calibration terminal can trigger local optimization correction under physical constraints to construct an objective function for parameter optimization. The objective function can be used to characterize the weighted sum of data fitting error and physical constraint violation penalty. The core of the objective function can be composed of the error square term of the measured output current and the model output current, the monotonicity penalty term of the series resistance, the boundary constraint penalty term, and the aging rate coefficient change rate penalty term. Each penalty term is matched with a pre-set penalty weight coefficient, and the sum of all weight coefficients is 1. The objective function can force the optimization process to meet the physical constraint rules through the penalty terms.

[0111] S502. Using the decoupled state parameter set of the two physics fields as the initial value of the iteration, search for the optimal solution that makes the objective function converge within the preset physical boundary threshold interval, and use the parameter set corresponding to the optimal solution as the correction parameter set.

[0112] Optionally, the physical boundary threshold range can be used to characterize the reasonable fluctuation range that each state parameter to be optimized can achieve at the physical level. The physical boundary threshold range can be preset based on the factory rated parameters of photovoltaic modules, the inherent physical characteristics of semiconductor devices and the full life cycle operation rules. The physical boundary threshold range can match the corresponding upper and lower limit thresholds for each parameter. The physical boundary threshold range can be used to limit the feasible domain of the optimization search, thereby avoiding the occurrence of physically unrealizable abnormal parameter values ​​during the optimization process.

[0113] For example, the intelligent correction terminal can use the aforementioned dual-physics decoupled state parameter set as the initial value for iterative optimization. The intelligent correction terminal can employ a constrained nonlinear programming optimization algorithm adapted to the online operating scenario. Within the feasible region defined by a preset physical boundary threshold interval, it conducts an iterative search with the objective function minimization as the optimization objective. During the iteration process, the candidate parameter values ​​at each step are strictly constrained to the upper and lower limit threshold ranges of the corresponding parameters. The intelligent correction terminal can determine in real time during the iteration process whether the objective function satisfies the preset convergence condition. The convergence condition may include, but is not limited to, the change in the objective function between adjacent iteration steps being less than the preset convergence threshold, or the number of iterations reaching the preset maximum number of iterations. When the objective function satisfies the preset convergence condition, the intelligent correction terminal can immediately stop the iteration and output the current optimal solution, using the parameter set corresponding to the optimal solution as the correction parameter set.

[0114] In this embodiment, the intelligent correction terminal uses the decoupled state parameter set of the two physical fields as the initial value for iteration. Within the physical boundary threshold range, it searches for the optimal solution of the objective function using a constrained nonlinear programming algorithm. This achieves local optimization of parameters under physical rationality constraints, ensuring that the correction parameter set strictly conforms to the physical characteristic boundaries of the photovoltaic device while meeting the data fitting accuracy requirements. This improves the reliability and engineering applicability of the state parameters.

[0115] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0116] Based on the same inventive concept, this application also provides a photovoltaic control parameter online calibration system for implementing the aforementioned photovoltaic control parameter online calibration method. The solution provided by this system is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the photovoltaic control parameter online calibration system provided below can be found in the limitations of the photovoltaic control parameter online calibration method described above, and will not be repeated here.

[0117] In one exemplary embodiment, such as Figure 2 As shown, a photovoltaic control parameter online correction system 600 is provided, comprising:

[0118] The parameter decoupling module 601 can be used to acquire real-time operating data of photovoltaic modules, input the real-time operating data into a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism, and obtain a dual-physics decoupling state parameter set and a state covariance matrix.

[0119] The optimization and correction module 602 can be used to perform physical consistency verification on each parameter of the dual-physics field decoupling state parameter set. When any parameter in the dual-physics field decoupling state parameter set violates the preset physical constraint rules, local optimization and correction under physical constraints is triggered to obtain the correction parameter set.

[0120] The predictive control module 603 can be used to acquire environmental data at the current moment, perform model predictive control calculations on the correction parameter set and environmental data, and obtain the theoretical maximum power point voltage; wherein, the theoretical maximum power point voltage is used to characterize the ideal operating voltage of the photovoltaic module when it outputs maximum power under the current environmental conditions;

[0121] The feedforward control module 604 can be used to generate a feedforward duty cycle command based on the theoretical maximum power point voltage; wherein, the feedforward duty cycle command is used to instruct the inverter to control the operating voltage of the photovoltaic module to approach the theoretical maximum power point voltage.

[0122] The feedback control module 605 can perform differential calculations on real-time operating data and environmental data to obtain the environmental change rate. Based on the state covariance matrix and the environmental change rate, it generates a feedback correction instruction and superimposes the feedforward duty cycle instruction and the feedback correction instruction to obtain the final control instruction. Among them, the feedback correction instruction is used to instruct the inverter to adjust the operating voltage of the photovoltaic module; the final control instruction is used to instruct the inverter to adjust the on-duty cycle of the switching transistor.

[0123] In one embodiment, the parameter decoupling module can also be used for:

[0124] Through the slow aging sub-model layer, port voltage, cell temperature and irradiance are extracted from real-time operating data to construct the input vector, output current is extracted from real-time operating data to construct the observation vector, and a state vector is defined; wherein, the state vector includes series resistance, aging rate coefficient, photocurrent and parallel resistance;

[0125] Through the rapidly changing environment sub-model layer, based on the cell temperature and irradiance, and combined with the preset mapping relationship between the photogenerated current, parallel resistance, and environmental factors, a first environmental estimate and a second environmental estimate are calculated; wherein, the first environmental estimate is used to characterize the environmental estimate of the photogenerated current; and the second environmental estimate is used to characterize the environmental estimate of the parallel resistance.

[0126] By using a dual-timescale extended Kalman filter prediction layer, state prediction is performed on the series resistance and aging rate coefficient to obtain a first predicted value and a second predicted value. The first predicted value, the second predicted value, the first environmental estimate, and the second environmental estimate are fused to obtain a predicted state vector. The first predicted value is used to characterize the predicted value of the series resistance, and the second predicted value is used to characterize the predicted value of the aging rate coefficient.

[0127] By extending the Kalman filter update layer through dual time scales, the predicted output current is obtained based on the predicted state vector and the input vector. The Kalman gain is obtained by calculating the residual between the predicted output current and the observation vector. The predicted state vector is corrected according to the Kalman gain to obtain the corrected state vector, and the state covariance matrix is ​​updated.

[0128] Through the output layer, the corrected state vector is output as the set of state parameters for decoupling the two physics fields, and the state covariance matrix is ​​output.

[0129] In one embodiment, the system may further include:

[0130] The offline parameter update module can be used to obtain the historical correction parameter set and the historical environmental data corresponding to the historical correction parameter set after physical consistency verification. The series resistance sequence in the historical correction parameter set is used as the dependent variable, and the cumulative irradiance and cumulative temperature stress in the historical environmental data are used as independent variables to construct an aging characteristic fitting model. The coefficients of the aging characteristic fitting model are solved by the offline regression algorithm to obtain the updated aging rate parameters.

[0131] The parameter synchronization module can be used to synchronize the updated aging rate parameters to the slow-varying aging sub-model layer in order to update the slow-varying aging sub-model layer.

[0132] In one embodiment, the feedback control module includes:

[0133] The index calculation unit can be used to extract diagonal elements from the state covariance matrix, construct the state estimation error variance vector, calculate the norm of the state estimation error variance vector, use the norm of the state estimation error variance vector as the state uncertainty index, normalize the environmental change rate and calculate the modulus value to obtain the environmental dynamic disturbance intensity index.

[0134] The gain adaptive mapping unit can be used to perform nonlinear mapping on the state uncertainty index and the environmental dynamic disturbance intensity index to obtain the feedback correction gain; wherein, the feedback correction gain is used to characterize the inverter's response intensity to voltage deviation.

[0135] The feedback command synthesis unit can be used to calculate the voltage tracking error between the theoretical maximum power point voltage and the real-time port voltage of the photovoltaic module. The product of the voltage tracking error and the feedback correction gain is used as the proportional correction term, and the environmental change rate is defined as the feedforward compensation term. The proportional correction term and the feedforward compensation term are linearly superimposed to obtain the feedback correction command.

[0136] In one embodiment, the optimization correction module includes:

[0137] The physical constraint verification unit can be used to perform multi-dimensional physical consistency verification on the decoupled state parameter set of dual physics fields, and obtain multiple verification results. When any of the multiple verification results is determined to be unsuccessful, local optimization correction under physical constraints is triggered to construct the objective function. The objective function is used to characterize the weighted sum of data fitting error and physical constraint violation penalty.

[0138] The objective function is expressed as follows:

[0139]

[0140] In the formula, Describe the objective function. This represents the measured output current. This represents the set of parameters to be optimized. Represents the set of parameters to be optimized. The output current of the model is obtained by substituting into the composite photovoltaic model. This represents the boundary constraint penalty term. This represents the monotonicity penalty term for the series resistance. This represents the penalty term for the rate of change of the aging rate coefficient. This represents the penalty weight coefficient, and ;

[0141] The parameter optimization and correction unit can be used to take the decoupled state parameter set of the two physics fields as the initial value of the iteration, search for the optimal solution that makes the objective function converge within the preset physical boundary threshold interval, and take the parameter set corresponding to the optimal solution as the correction parameter set.

[0142] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the online correction method for photovoltaic control parameters as described above.

[0143] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0144] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0145] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for online calibration of photovoltaic control parameters, characterized in that, The method includes: S101. Obtain real-time operating data of photovoltaic modules, input the real-time operating data into a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism, and obtain a dual-physics field decoupled state parameter set and a state covariance matrix. S102. Perform physical consistency verification on each parameter of the dual-physics field decoupling state parameter set. When any parameter in the dual-physics field decoupling state parameter set violates the preset physical constraint rules, trigger local optimization correction under physical constraints to obtain the correction parameter set. S103. Obtain the environmental data at the current moment, perform model prediction control calculation on the correction parameter set and the environmental data to obtain the theoretical maximum power point voltage; wherein, the theoretical maximum power point voltage is used to characterize the ideal operating voltage of the photovoltaic module when it outputs maximum power under the current environmental conditions; S104. Based on the theoretical maximum power point voltage, generate a feedforward duty cycle command; wherein, the feedforward duty cycle command is used to instruct the inverter to control the operating voltage of the photovoltaic module to approach the theoretical maximum power point voltage; S105. Perform differential operations on the real-time operating data and the environmental data to obtain the environmental change rate. Based on the state covariance matrix and the environmental change rate, generate a feedback correction instruction and superimpose the feedforward duty cycle instruction and the feedback correction instruction to obtain the final control instruction. The feedback correction instruction is used to instruct the inverter to adjust the operating voltage of the photovoltaic module. The final control instruction is used to instruct the inverter to adjust the on-duty cycle of the switching transistor.

2. The method according to claim 1, characterized in that, The composite photovoltaic model includes a slow-changing aging sub-model layer, a fast-changing environment sub-model layer, a dual-timescale extended Kalman filter prediction layer, a dual-timescale extended Kalman filter update layer, and an output layer. The real-time operating data is input into a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism to obtain a dual-physics decoupled state parameter set and a state covariance matrix, including: S201. Through the slow-changing aging sub-model layer, the port voltage, cell temperature and irradiance are extracted from the real-time operating data to construct the input vector, the output current is extracted from the real-time operating data to construct the observation vector, and the state vector is defined; wherein, the state vector includes series resistance, aging rate coefficient, photocurrent and parallel resistance; S202. Through the rapidly changing environment sub-model layer, based on the cell temperature and the irradiance, and combined with the preset mapping relationship between the photocurrent, parallel resistance, and environmental factors, a first environmental estimate and a second environmental estimate are calculated; wherein, the first environmental estimate is used to characterize the environmental estimate of the photocurrent; and the second environmental estimate is used to characterize the environmental estimate of the parallel resistance. S203. Through the dual-timescale extended Kalman filter prediction layer, state prediction is performed on the series resistance and the aging rate coefficient to obtain a first predicted value and a second predicted value. The first predicted value, the second predicted value, the first environmental estimate, and the second environmental estimate are fused to obtain a predicted state vector. The first predicted value is used to characterize the predicted value of the series resistance, and the second predicted value is used to characterize the predicted value of the aging rate coefficient. S204. Through the dual-timescale extended Kalman filter update layer, based on the predicted state vector and the input vector, the predicted output current is obtained. By calculating the residual between the predicted output current and the observation vector, the Kalman gain is obtained. The predicted state vector is corrected according to the Kalman gain to obtain the corrected state vector, and the state covariance matrix is ​​updated. S205. Through the output layer, output the modified state vector as the set of decoupled state parameters of the dual physics field, and output the state covariance matrix.

3. The method according to claim 2, characterized in that, The method further includes: S301. Obtain the historical correction parameter set that has passed the physical consistency verification and the historical environmental data corresponding to the historical correction parameter set. Take the series resistance sequence in the historical correction parameter set as the dependent variable and the cumulative irradiance and cumulative temperature stress in the historical environmental data as independent variables to construct an aging characteristic fitting model. Solve the coefficients of the aging characteristic fitting model through an offline regression algorithm to obtain the updated aging rate parameters. S302. Synchronize the updated aging rate parameter to the slow-changing aging sub-model layer to update the slow-changing aging sub-model layer.

4. The method according to claim 1, characterized in that, The step of generating feedback correction instructions based on the state covariance matrix and the environmental change rate includes: S401. Extract diagonal elements from the state covariance matrix to construct a state estimation error variance vector, calculate the norm of the state estimation error variance vector, use the norm of the state estimation error variance vector as a state uncertainty index, normalize the environmental change rate and calculate the modulus to obtain an environmental dynamic disturbance intensity index. S402. Perform a nonlinear mapping between the state uncertainty index and the environmental dynamic disturbance intensity index to obtain the feedback correction gain; wherein, the feedback correction gain is used to characterize the response intensity of the inverter to voltage deviation; S403. Calculate the voltage tracking error between the theoretical maximum power point voltage and the real-time port voltage of the photovoltaic module. Use the product of the voltage tracking error and the feedback correction gain as a proportional correction term, and define the environmental change rate as a feedforward compensation term. Linearly superimpose the proportional correction term and the feedforward compensation term to obtain the feedback correction command.

5. The method according to claim 1, characterized in that, The physical consistency of each parameter in the dual-physics decoupling state parameter set is verified. When any parameter in the dual-physics decoupling state parameter set violates a preset physical constraint rule, local optimization correction under physical constraints is triggered to obtain a correction parameter set, including: S501. Perform multi-dimensional physical consistency verification on the dual-physics field decoupled state parameter set to obtain multiple verification results. When any of the multiple verification results is determined to be unsuccessful, trigger local optimization correction under the physical constraints to construct an objective function. The objective function is used to characterize the weighted sum of data fitting error and physical constraint violation penalty. The expression for the objective function is: In the formula, Denotes the objective function, This represents the measured output current. This represents the set of parameters to be optimized. This represents the set of parameters to be optimized. Substituting the values ​​into the composite photovoltaic model, the calculated model output current... This represents the boundary constraint penalty term. This represents the monotonicity penalty term for the series resistance. This represents the penalty term for the rate of change of the aging rate coefficient. This represents the penalty weight coefficient, and ; S502. Using the dual-physics decoupling state parameter set as the initial value for iteration, search for the optimal solution that makes the objective function converge within the preset physical boundary threshold interval, and use the parameter set corresponding to the optimal solution as the correction parameter set.

6. A photovoltaic control parameter online correction system, characterized in that, The system includes: The parameter decoupling module is used to acquire real-time operating data of photovoltaic modules, input the real-time operating data into a composite photovoltaic model based on a dual-time-scale extended Kalman filter mechanism, and obtain a dual-physics field decoupled state parameter set and a state covariance matrix. The optimization and correction module is used to perform physical consistency verification on each parameter of the dual-physics field decoupling state parameter set. When any parameter in the dual-physics field decoupling state parameter set violates the preset physical constraint rules, local optimization and correction under physical constraints is triggered to obtain the correction parameter set. The predictive control module is used to acquire environmental data at the current moment, perform model predictive control calculations on the correction parameter set and the environmental data, and obtain the theoretical maximum power point voltage; wherein, the theoretical maximum power point voltage is used to characterize the ideal operating voltage of the photovoltaic module when it outputs maximum power under the current environmental conditions; A feedforward control module is used to generate a feedforward duty cycle command based on the theoretical maximum power point voltage; wherein, the feedforward duty cycle command is used to instruct the inverter to control the operating voltage of the photovoltaic module to approach the theoretical maximum power point voltage; The feedback control module is used to perform differential operations on the real-time operating data and the environmental data to obtain the environmental change rate. Based on the state covariance matrix and the environmental change rate, it generates a feedback correction instruction and superimposes the feedforward duty cycle instruction and the feedback correction instruction to obtain the final control instruction. The feedback correction instruction is used to instruct the inverter to adjust the operating voltage of the photovoltaic module. The final control instruction is used to instruct the inverter to adjust the on-duty cycle of the switching transistor.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images